Computational study on subfamilies of piperidine derivatives: QSAR modelling, model external verification, the inter-subset similarity determination, and structure-based drug designing

ABSTRACT

A new subset of furan-pyrazole piperidine derivatives was used for QSAR model development. These compounds exhibit good Akt1 inhibitory activity; moreover, antiproliferative activities in vitro against OVCAR-8 (Human ovarian carcinoma cells) and HCT116 (human colon cancer cells), were confirmed for them. Based on the relevant three-dimensional (3D) and 2D autocorrelation descriptors, selected by genetic algorithm (GA), multiple linear regression (MLR) was established on half maximal-inhibitory concentration (IC₅₀), in Akt1 and cancer cell lines independently. Robustness, stability, and predictive ability of the models were evaluated using external and internal validation (r²: 0.742–0.832, Q²_LOO: 0.684–0.796, RMSE: 0.247–0.299, F: 32.283–57.578, and r²_y-random: 0.049–0.080). Furthermore, in the new strategy, each of the evaluated models was generalized to two other subfamilies of piperidines to simultaneously compare the activities and structural similarity of these three subsets. Probably, structural similarity can be more considered as a criterion of similarity in the mechanism of action. Also, external verification of suggested predictive models was performed by another subset. Finally, by focusing on M64 as the most potent in vivo antitumor compound, 15 new derivatives were designed and six potent candidates were proposed for further investigation.

KEYWORDS:

• GA-MLR
• furan-pyrazole piperidines
• three-dimensional descriptors
• conformational restriction
• Williams plot-based analyses
• structure-based drug design

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed at: https://doi.org/10.1080/1062936X.2021.1891568.